A fuel cell converts chemical energy into electrical energy and some thermal energy by means of a chemical reaction between a fuel (e.g., hydrogen gas or a hydrogen-containing fluid) and an oxidant (e.g., oxygen). A proton exchange membrane (PEM) fuel cell uses hydrogen or hydrogen-rich reformed gases as the fuel, a direct-methanol fuel cell (DMFC) uses methanol-water solution as the fuel, and a direct ethanol fuel cell (DEFC) uses ethanol-water solution as the fuel, etc. These types of fuel cells that require utilization of a PEM layer as a proton transport electrolyte are collectively referred to as PEM-type fuel cells.
A PEM-type fuel cell is typically composed of a seven-layered structure, including (a) a central PEM electrolyte layer for proton transport; (b) two electro-catalyst layers on the two opposite primary surfaces of the electrolyte membrane; (c) two fuel or gas diffusion electrodes (GDEs, hereinafter also referred to as diffusers) or backing layers stacked on the corresponding electro-catalyst layers (each GDE comprising porous carbon paper or cloth through which reactants and reaction products diffuse in and out of the cell); and (d) two flow field plates (or a bi-polar plate) stacked on the GDEs. The flow field plates are typically made of graphite, metal, or conducting composite materials, which also serve as current collectors. Gas-guiding channels are defined on a GDE facing a flow field plate or, more typically, on a flow field plate surface facing a GDE. Reactants (e.g., H2 or methanol solution) and reaction products (e.g., CO2 at the anode of a DMFC, and water at the cathode side) are guided to flow into or out of the cell through the flow field plates. The configuration mentioned above forms a basic fuel cell unit. Conventionally, a fuel cell stack comprises a number of basic fuel cell units that are electrically connected in series to provide a desired output voltage. If desired, cooling channels and humidifying plates may be added to assist in the operation of a fuel cell stack.
In one common practice, a fuel flow field plate and an oxidant gas flow field plate are separately made and then assembled together to form a bipolar plate (one side of a bipolar plate serving as a negative terminal and the other side as a positive terminal, hence the name). In some cases, an additional separator is sandwiched between the two flow field plates to form a bipolar plate. It would be highly advantageous if the flow filed plates and the separator can be mass-produced into an integrated bipolar plate assembly. This could significantly reduce the overall fuel cell production costs and reduce contact ohmic losses across constituent plate interfaces. The bipolar plate is known to significantly impact the performance, durability, and cost of a fuel cell system. The bipolar plate, which is typically machined from graphite, is one of the most costly components in a PEM fuel cell.
Fluid flow field plates have open-faced channels formed in one or both opposing major surfaces for distributing reactants to the gas diffuser plates, which are the anode and cathode backing layers, typically made of carbon paper or fabric. The open-faced channels also provide passages for the removal of reaction products and depleted reactant streams. Optionally, a bipolar plate may have coolant channels to manage the fuel cell temperature. According to the US Department of Energy (DOE), a bipolar plate should have the following desirable characteristics: high electrical conductivity (e.g., preferably having a thickness-direction conductivity no less than 100 S/cm and specific areal conductivity no less than 200 S/cm2), low permeability to fuel or oxidant fluids, good corrosion resistance, and good structural integrity. The specific areal conductivity is essentially the bipolar plate thickness-direction conductivity divided by the plate thickness. Hence, it is highly desirable to have a thinner plate. Current graphite bipolar plates, typically 3-5 mm thick, should preferably be reduced to below 1 mm and most preferably below 0.5 mm.
Conventional methods of fabricating fluid flow field plates require the engraving or milling of flow channels into the surface of rigid plates formed of a metal, graphite, or carbon-resin composite. Such plates are expensive due to high machining costs. The machining of channels into the graphite plate surfaces causes significant tool wear and requires significant processing times. Metals can be readily shaped into very thin plates, but long-term corrosion is a major concern. A corrosion-resistant coating may be used, but it has to be applied perfectly. The coating may also increase contact resistance.
Alternatively, fluid flow field plates can be made by a lamination process (e.g., U.S. Pat. No. 5,300,370, issued Apr. 5, 1994), wherein an electrically conductive, fluid impermeable separator layer and an electrically conductive stencil layer are consolidated to form one open-faced channel. Presumably, two conductive stencil layers and one separator layer may be laminated to form a bipolar plate. It is often difficult and time-consuming to properly position and align the separator and stencil layers. Die-cutting of stencil layers require a minimum layer thickness, which limits the extent to which fuel cell stack thickness can be reduced. Such laminated fluid flow field assemblies tend to have higher manufacturing costs than integrated plates, due to the number of manufacturing steps associated with forming and consolidating the separate layers. They are also prone to delamination due to poor interfacial adhesion and vastly different coefficients of thermal expansion between a stencil layer (typically a metal) and a separator layer. Corrosion also presents a challenging issue for metal-based bipolar plates in a PEM fuel cell since they are used in an acidic environment.
A variety of composite bipolar plates have been developed, which are mostly made by compression molding of polymer matrices (thermoplastic or thermoset resins) filled with conductive particles such as graphite powders or fibers. Because most polymers have extremely low electronic conductivity, excessive conductive fillers have to be incorporated, resulting in an extremely high viscosity of the filled polymer melt or liquid resin and, hence, making it very difficult to process. Bi-polar plates for use in PEM fuel cells constructed of graphite powder/fiber filled resin composite materials and having gas flow channels are reviewed by Wilson, et al (U.S. Pat. No. 6,248,467, Jun. 19, 2001). Injection-molded composite-based bipolar plates are disclosed by Saito, et al. (U.S. Pat. No. 6,881,512, Apr. 19, 2005 and U.S. Pat. No. 6,939,638, Sep. 6, 2005). These thermoplastic or thermoset composites exhibit a bulk conductivity significantly lower than 100 S/cm (the US Department of Energy target value), typically not much higher than 10 S/cm.
Besmann, et al. disclosed a carbon/carbon composite-based bipolar plate (U.S. Pat. No. 6,171,720 (Jan. 9, 2001) and U.S. Pat. No. 6,037,073 (Mar. 14, 2000)). The manufacture process consists of multiple steps, including production of a carbon fiber/phenolic resin preform via slurry molding, followed by a compression-molding step. The molded part is then pyrolyzed at a high temperature (1,500° C.-2,500° C.) to obtain a highly porous carbon/carbon composite. This is followed by chemical vapor infiltration (CVI) of a carbon matrix into this porous structure. It is well-known that CVI is a very time-consuming and energy-intensive process and the resulting carbon/carbon composite, although exhibiting a high electrical conductivity, is very expensive.
Instead of using pyrolyzation and CVI to produce carbon/carbon composites, Huang, et al. (US Patent Application Pub. No. 2004/0229993, Nov. 18, 2004) discloses a process to produce a thermoplastic composite with a high graphite loading. First, polymer fibers, such as thermotropic liquid crystalline polymers or polyester, reinforcing fibers such as glass fibers, and graphite particles are combined with water to form a slurry. The slurry is pumped and deposited onto a sieve screen. The sieve screen serves the function of separating the water from the mixture of polymer fibers, glass fibers and graphite. The mixture forms a wet-lay sheet which is placed in an oven. Upon heating to a temperature sufficient to melt the polymer fibers, the wet-lay sheet is allowed to cool and have the polymer material solidify. Upon solidification, the wet-lay sheet takes the form of a sheet material with reinforcement glass fibers held together by globules of thermoplastic material, and graphite particles adhered to the sheet material by the thermoplastic material. Several of these sheets are then stacked, preferably with additional graphite powder interspersed between sheets, and compression-molded in a hot press. After application of heat and pressure in the press, one or more formed bipolar plates are obtained, where the bipolar plates are a composite of glass fibers, thermoplastic matrix and graphite particles. Clearly, this is also a tedious process which is not amenable to mass production.
Alternatively, fluid flow field plates can be made from an electrically conductive, substantially fluid impermeable material that is sufficiently compressible or moldable so as to permit embossing. Flexible graphite sheet is generally suitable for this purpose because it is relatively impervious to typical fuel cell reactants and coolants and thus is capable of isolating the fuel, oxidant, and coolant fluid streams from each other. It is also compressible and embossing processes may be used to form channels in one or both major surfaces. The “flexible graphite” is typically obtained in the following manner: Natural graphite particles are treated with an agent that intercalates into the graphite crystal structure (inter-graphene layer spaces) to form a graphite intercalation compound (GIC) or “expandable graphite.” Rapid heating of a GIC or expandable graphite to a high temperature, typically 700-1,050° C., results in a large expansion of the graphite crystal structure by typically 80-300 times in the c-axis direction, the direction that is perpendicular to the graphene plane or basal plane of the graphite crystal structure. The exfoliated graphite particles are vermiform in appearance, and are therefore commonly referred to as graphite worms. Hereinafter, the term “exfoliated graphite” will be used interchangeably with the term “expanded graphite” or graphite worms. The worms are typically characterized as having exfoliated flakes that are substantially interconnected. An “exfoliated flake” is typically composed of one or multiple graphene planes (sheets) bonded together by van der Waals forces with an inter-planar spacing of typically from 0.335 (graphite) to 0.6 nm (graphite oxide) between two un-expanded basal planes inside a flake. However, there are pores between exfoliated flakes that are typically between 10 nm and 10 μm wide. These pores make the worms fluffy and compressible. The worms may be re-compressed together into flexible sheets which, unlike the original graphite flakes, can be easily formed and cut into various shapes. These thin sheets (foils or films) are hereinafter referred to as flexible graphite. Flexible graphite can be wound up on a drum to form a roll of thin film, just like a roll of thin plastic film or paper.
Although flexible graphite sheets are highly conductive (in a direction parallel to the sheet plane), they by themselves may not have sufficient stiffness and must be supported by a core layer or impregnated with a resin. For example, Wilkinson, et al., in U.S. Pat. No. 5,527,363 (Jun. 18, 1996), disclosed a fluid flow field plate comprising a metal sheet interposed between two flexible graphite (FG) sheets having flow channels embossed on a major surface thereof. Prior art flexible graphite sheets typically have a thickness-direction conductivity up to only 15 S/cm, although its in-plane conductivity may be greater than 1,300 S/cm. These FG-metal-FG laminates are expected to exhibit a thickness-direction conductivity less than 100 S/cm, the US DOE requirement. This may be illustrated as follows: Assume that the top layer, bottom layer, and core layer of the three-layer laminate all have a thickness of 0.15 mm (150 μm) and that the core layer is a conducting metal foil having a conductivity of 5×105 S/cm. The three layers may be considered as being connected in series electrically. Then, a simple calculation would predict that the thickness-direction conductivity of the resulting laminate is approximately 22.5 S/cm, lower than the DOE requirement. FIG. 1 shows the thickness-direction conductivity of the laminate plotted as a function of the thickness-direction conductivity of flexible graphite layers. The diagram indicates that the thickness-direction conductivity of the three-layer laminate will exceed 100 S/cm if the FG layers have a thickness-direction conductivity greater than 67 S/cm. Prior art flexible graphite sheets fall short of this conductivity level.
Alternatively, Mercuri, et al. (e.g., U.S. Pat. No. 5,885,728, Mar. 23, 1999 and U.S. Pat. No. 6,037,074, Mar. 14, 2000) disclosed a flexible graphite sheet having embedded ceramic or glass fibers extending from its surface into the sheet to increase the resin permeability of the sheet for the preparation of a resin-impregnated flexible graphite bipolar plate. By allowing ceramic or glass fibers to puncture through layers of exfoliated graphite also leave these layers vulnerable to gas permeation, thereby significantly reducing the hydrogen and oxygen permeation resistance of a bipolar plate and increasing the chance of dangerous mixing of hydrogen and oxygen inside a fuel cell stack.
What follows is a summary of the state of the art of the flexible graphite sheet, resin-impregnated expanded graphite composite, resin-impregnated flexible graphite sheet composite, and methods of producing these materials:
Olstowski, et al. (“Novel Compressed Cohered Graphite Structures and Method of Preparing Same,” U.S. Pat. No. 3,492,197, Jan. 27, 1970) provided compressed and resin-bonded forms of expanded vermicular graphite. The resin-bonded composite is obtained by (a) providing a supply of an expanded vermicular graphite having an apparent bulk density of 0.2-2.0 pounds per cubic foot; (b) providing a supply of a bonding agent; (c) blending the expanded vermicular graphite and bonding agent in an amount of 2-35 weight percent bonding agent based on the total weight of the expanded graphite-bonding agent mixture; (d) compressing the mixture at a pressure of 5-50,000 psi in predetermined directions into predetermined forms of cohered graphite; and (e) treating the so-formed composite to activate the bonding agent thereby promoting adhesion within the compact. This invention taught about compressing vermicular-bonding agent mixture in a uniaxial direction to produce a highly anisotropic composite and in bi-axial, tri-axial, cylinder-radial, and isostatic directions to produce less anisotropic or more isotropic composites. However, it failed to teach, implicitly or explicitly, how a desired degree of isotropy could be maintained when the bi-axially, tri-axially, cylinder-radially, and isostatically compressed composite compacts (prior to curing or fusing to consolidate) were re-compressed or molded as a final operation to become a thin composite plate. This thin plate (thinner than 5 mm, preferably thinner than 3 mm, further preferably thinner than 1 mm, and most preferably thinner than 0.5 mm) is for a bipolar plate application. Further, this patent was limited to using a solid bonding agent to begin with the blending process, excluding liquid polymers from the invention due to the perceived notion that these liquid polymers “can prevent formation of highly densified composites.” This patent did not teach how bi-axial, tri-axial, cylinder-radial, and isostatic compressions could be accomplished in a real manufacturing environment for the mass production of less anisotropic composites on a continuous basis. Furthermore, the method disclosed in this patent entailed first exfoliating graphite to obtain graphite worms and then mixing graphite worms with a bonding agent in a fine solid powder form. Once the graphite worms are formed, it would be very difficult to mix the worms with fine solid particles in a homogeneous manner without breaking up or significantly disturbing the continuous network of electron-transport paths (interconnected graphite flakes).
Caines (“Vermicular Expanded Graphite Composite Materials,” U.S. Pat. No. 4,265,952, May 5, 1981) disclosed an expanded graphite composite containing a corrosion resistant resin (e.g., polytetrafluoroethylene, PTFE). The composite was prepared by blending vermicular graphite with a suspension of fine solid resin particles in a carrier liquid medium, vaporizing the carrier, and heating the composite material to sinter the resin. No electrical property of the resulting composite was reported.
Atkinson, et al. (“Housing for Electrical or Electronic Equipment,” U.S. Pat. No. 4,530,949, Jul. 23, 1985) provided a low-density composite composition consisting of exfoliated graphite and a thermosetting resin binder. The density (<0.1 gm/cm3) and the electrical conductivity (0.1 S/cm) values are relatively low.
Fukuda, et al. (“Reinforced Flexible Graphite Sheet,” U.S. Pat. No. 4,729,910, Mar. 8, 1988) disclosed a process of producing thermosetting resin reinforced flexible graphite sheets. The process involved subjecting both the flexible graphite sheet and a phenolic resin solution to a preliminary de-aeration treatment prior to immersing the flexible graphite sheet in the resin solution. No electrical conductivity data was offered.
Chung provided a low-density (0.7 gm/cm3) exfoliated flexible graphite flake-reinforced composite with a conductivity of 2 S/cm (Chung, “Low-Density Graphite-Polymer Electrical Conductor,” U.S. Pat. No. 4,704,231, Nov. 3, 1987). Chung also provided an in-situ exfoliation method of producing graphite flake-reinforced epoxy composites (“Composites of In-Situ Exfoliated Graphite,” U.S. Pat. No. 4,946,892, Aug. 7, 1990).
Fong, et al. (“Methacrylate Impregnated Carbonaceous Parts,” U.S. patent application Ser. No. 09/896,178, filed on Jun. 29, 2001 (Pub. No. US 2001/0046560, Pub date Nov. 29, 2001)) disclosed a method of impregnating a highly porous carbon material with a methacrylate polymer. No electrical conductivity data was provided.
Öttinger, et al. (“Impregnated Bodies Made of Expanded Graphite, Process for Producing Such Bodies and Sealing Elements, Fuel Cell Components and Heat-Conducting Elements Formed of the Bodies,” U.S. Pat. No. 6,746,771, Jun. 8, 2004) provided composites of expanded graphite impregnated with isocyanate or epoxy resins. The method involved soaking expanded graphite with a low-viscosity, polymerizing resin. The achievable electrical conductivity of the resulting composites appears to be in the range of 2-10 S/cm.
Da Silva, et al. (“Method for Producing Composite Objects Using Expanded Graphite and Vermiculite,” U.S. patent application Ser. No. 10/574,803 filed on Oct. 8, 2004 (Pub. No. US 2007/0015267, Pub date Jan. 18, 2007)) disclosed a method of producing s composite object consisting of at least two distinct parts.
Mercuri, et al. (“Flexible Graphite Article and Method of Manufacture,” U.S. Pat. No. 6,432,336, Aug. 13, 2002 and U.S. Pat. No. 6,706,400, Mar. 16, 2004) disclosed a resin-impregnated flexible graphite sheet exhibiting enhanced isotropy and a method of producing resin-impregnated flexible graphite sheet. The method includes the steps of (i) reacting raw natural graphite flake-like particles with a liquid intercalant solution to form intercalated graphite particles; (ii) exposing the intercalated graphite particles to a temperature of at least about 700° C. to expand the intercalated graphite particles to form a stream of exfoliated graphite particles; (iii) continuously compressing the stream of exfoliated graphite particles into a continuous coherent self-supporting mat of flexible graphite; (iv) continuously contacting the flexible graphite mat with liquid resin and impregnating the mat with liquid resin; and (v) continuously calendering the flexible graphite mat to increase the density thereof to form a continuous flexible graphite sheet having a thickness of no more than about 1.0 inch.
It is of interest to note that this process disclosed by Mercuri, et al. involves compressing the exfoliated graphite into a flat mat prior to impregnating the mat with a resin. This sequence is disadvantageous in that the re-compressed flexible graphite, being much denser, is less permeable to resin impregnation. Furthermore, uniaxial re-compression of the exfoliated graphite prior to resin impregnation tends to align or orientate the graphite flakes along the graphite sheet plane direction (perpendicular to the re-compression vector), resulting in a more anisotropic flexible graphite sheet composite. Once these flakes were well-aligned in a sheet to form a highly cohered mat, their orientations could no longer be changed during subsequent resin impregnation and molding operations. Furthermore, no attempt was made to re-compress the mat in different directions. Thin graphite flakes are essentially single crystals with the flake plane parallel to the basal plane and, hence, exhibit a high electrical conductivity along thin flake plane directions and much lower conductivity along the thickness direction, or c-axis direction. Consequently, the bipolar plates prepared by using the Mercuri process are not expected to have a high thickness-direction conductivity.
The resin-impregnated flexible graphite sheet exhibiting enhanced isotropy as disclosed by Mercuri, et al. (U.S. Pat. No. 6,706,400) was said to contain interlocked particles of expanded graphite. A portion of these interlocked particles of expanded graphite was substantially unaligned with the opposed planar surfaces. However, Mercuri, et al. did not fairly specify how these unaligned graphite flakes were obtained. Presumably, this could be achieved by mixing large particles of exfoliated graphite with smaller particles of exfoliated graphite, as implied in a Mercuri's earlier patent (U.S. Pat. No. 5,846,459, Dec. 8, 1998). The trade literature published by GrafTech (assignee of Mercuri's patents) indicates the electrical resistivity of bipolar plates in the X-Y plane as 7 μOhm-m (in-plane conductivity=1428 S/cm) and in the Z-direction as 300 μOhm-m (thickness-direction conductivity=33 S/cm). The thickness-direction conductivity is unsatisfactory.
In addition to exhibiting high electrical conductivity, the flow field plate or bipolar plate should be constructed from inexpensive starting materials, materials that are easily formed into any plate configuration, preferably using a continuous molding process, and materials that are corrosion resistant in low temperature fuel cells and that do not require further processing such as high temperature pyrolyzation treatments. The above review clearly indicates that prior art bipolar plate material compositions and processes have not provided a satisfactory solution for the fuel cell industry.
In our earlier applications, we disclosed a sheet molding compound (SMC) composition particularly for use as a fuel cell flow field plate or bipolar plate [Bor Z. Jang, “Sheet Molding Compound Flow Field Plate, Bipolar Plate and Fuel Cell,” U.S. patent application Ser. No. 11/293,540 (Dec. 5, 2005) and Bor Z. Jang, A. Zhamu, Lulu Song, “Method for Producing Highly Conductive Sheet Molding Compound, Fuel cell Flow Field Plate, and Bipolar Plate,” U.S. patent application Ser. No. 11/293,541 (Dec. 5, 2005)]. In one preferred embodiment, the SMC composition comprises a top FG sheet, a bottom FG sheet, and a nano filler-resin mixture sandwiched between the top sheet and the bottom sheet. The flexible graphite sheet has a planar outer surface having formed therein a fluid flow channel. The nano filler-resin mixture comprises a thermoset resin and a conductive nano filler (e.g., nano graphene plates or graphitic nano fibers) present in a sufficient quantity to render the SMC composition electrically conductive enough to be a current collector material. When the resin is cured or solidified, the two sheets are well bonded by the resin to provide good structural integrity to the resulting “laminated” structure.
Again, assume that the top layer, bottom layer, and core layer of the three-layer laminate all have a thickness of 0.15 mm (150 μm) and that the core layer is a conducting nanocomposite having a conductivity of 100 S/cm. The three layers may be considered as being connected in series electrically. FIG. 2 shows the thickness-direction conductivity of the laminated SMC plotted as a function of the thickness-direction conductivity of the flexible graphite layers. The diagram indicates that the thickness-direction conductivity of the SMC will exceed 100 S/cm if the FG layers have a thickness-direction conductivity greater than 100 S/cm. Hence, it is highly desirable to have flexible graphite sheets having a high thickness-direction conductivity. However, conventional flexible graphite normally has a thickness-direction conductivity less than 15 S/cm.
Accordingly, an object of the present invention is to provide an exfoliated graphite composition that exhibits a relatively high thickness-direction conductivity, preferably greater than 35 S/cm, more preferably greater than 67 S/cm, most preferably greater than 100 S/cm.
Another object of the present invention is to provide an exfoliated graphite composition that can be easily molded or embossed into a flow field plate, bipolar plate, or current collector.
Still another object of the present invention is to provide an exfoliated graphite article that is intrinsically less anisotropic.
Yet another object of the present invention is to provide a process for producing exfoliated graphite articles with enhanced isotropy. Such a process can be continuous, automated, and adaptable for mass production of bipolar plates.